← ClaudeAtlas

researchzy-systematic-reviewlisted

Run a PRISMA 2020 systematic review: coded protocol, logged searches of Researchzy, PubMed, ClinicalTrials.gov and user exports, deduplication, user-calibrated screening with quoted reasons, extraction table, and flow counts from a per-record ledger. Use for a systematic review of one question.
Researchzy/Researchzy-academic-skills · ★ 0 · Code & Development · score 72
Install: claude install-skill Researchzy/Researchzy-academic-skills
# Researchzy Systematic Review This skill takes one question to a PRISMA 2020 flow diagram and report: a protocol with coded criteria, searches of several sources each logged with its date and counts, deduplication, title-abstract and full-text screening calibrated against the user's own decisions, an extraction table, and the flow counts. You search, read and judge; `sr_ledger.py` keeps one entry per record, and `sr_prisma.py` computes every number of the flow from those entries, so no count is typed by hand. It does not claim that a search is complete, assess risk of bias (`researchzy-risk-of-bias`), compute or pool effects (`researchzy-meta-analysis`; both in plugin `researchzy-evidence`, when installed), or write narrative and scoping reviews (`researchzy-literature-review`, plugin `researchzy`). It never logs in to subscription databases, reads registry web pages or formats Word files. ## Requirements - Python 3.10 or newer. The three scripts need nothing else and never use the network: you fetch and save the raw results, the scripts read the files. Resolve `SKILL_DIR` to the directory containing this `SKILL.md`. - The Researchzy tools (`search_papers`, `get_pdf_download`) come with the `researchzy` plugin and need a Researchzy account; they search the Researchzy index and fetch full texts. Without them, say that the index cannot be searched, and continue with PubMed, ClinicalTrials.gov and the user's exports only if the user agrees in step 1; the limitations then sa